- Dead code from experiments in SelectionBar removed.
- Many warnings about unused parameters fixed with WXUNUSED()
- Many warnings about signed / unsigned comparisons cleaned up.
- Several 'local variable declared but not used' warnings fixed.
The toggle to switch the extra menu(s) on or off still says 'menus' (plural). Decided not to change it as the submenus count as 'menus'. Also we may later add extra menus to the track drop down and in other places.
Currently, when Audacity exports labels tracks numbers are formatted
with the "%f" specifier, which relies on the current user locale to pick
up the decimal separator.
This means that the numbers indicating the start and the end of the
labeled interval could end up using a comma as a decimal separator.
This makes the exported file less portable, especially in the case of
parsing the exported file with an external tool.
Audacity is already able to _import_ labels tracks independently of the
locale because it uses Internat::CompatibleToDouble(), so do something
symmetric at _export_ time, using Internat::ToString() to make the
exported data locale-independent (i.e. always using a dot as the decimal
separator).
Proposed in https://sourceforge.net/p/audacity/mailman/message/35534945/
NOTE:
When converting numbers to strings FLT_DIG is passed as the
digitsAfterDecimalPoint argument of Internat::ToString(), this is in
order to preserve the look and the alignment of the previous "%f" format
specifier; without that Internat::ToString() would strip trailing zeros.
Audacity requires C++11, this ensures that FLT_DIG is defined.
... "#trebmuh" comment lines remain, where the translator may wish to review
and revise later.
Olivier's commit comments:
- disambiguation: decay (EN) -> déclin (FR)
- disambiguation : save (EN) -> sauvegarder (FR) and record (EN) -> enregistrer (FR)
- consistency around preset (EN) -> préréglage (FR)
- fixes "sample rate" which was translated with "fréquence d'échantillonnage" while the correct translation is "taux d'échantillonnage"
- fixes "bit rate" which was translated with "débit" while the correct translation is "débit binaire"
- fixes a misleading translation where a "bit rate" (EN) was translated to "taux d'échantillonnage" instead of "débit binaire"
+ one string was half-translated
+ 5 typos
+ 1 frenglish
+ 1 better translation
+ 3 comments-to-self
+ updated timestamp
+ 1 missing translation
adds Christian Brochec in the header
+ a bunch of note-to-self precisions
+ a couple of better accelerators
+ typos
+ update the timestamp in the header
+ get the rid off the (useless now) "#| msgid" (which were ancient translation)
... but there are still 1029 errors from
msgcmp fr.po audacity.pot
Many of which could be fixed by removing the #, fuzzy comment lines, if
translator believes the translation is correct.
... Easily done with the following command.
ls *.po | xargs -t -n1 -I {} msgmerge -U --previous --add-location=file {} audacity.pot
Resulting files will have all the blanks to be filled in by the translators for
new messages.
--previous keeps the old msgids in comments, for which matches were only fuzzy,
as an aid to the translator.
--add-location=file includes file names only in comments, omitting line
numbers. If done again next time the files merge, then this will lessen the
diffs.
The results may include commented-out obsolete messages at the end, which
start with #~
At translator's discretion, these may be discarded when no longer helpful with
this command, substituting your filename twice for xx.po:
msgattrib -o xx.po --no-obsolete xx.po
... Such projects as described in
http://forum.audacityteam.org/viewtopic.php?f=47&t=97787
(That thread did not describe the crashes, but I came across them, debugging,
or deduced the possibility of them)
In which all the shared_ptrs to BlockFile objects ended up pointing at one
common file; but the size of this file was sometimes larger than the
Sequence's maximum block size, or not the same as the difference between one
block's start offset and the next one's.
